In a trio of announcements this week, independent object storage provider MinIO is making it possible to implement cloud-native storage without requiring its customers to know Kubernetes (K8s). And, spurred on by its roughly six-month old partnership with VMware, it is expanding its live production support with a health check option and a new self-service console that avoids the need to get health stats via APIs.
The common thread of the announcements is enabling customers to implement MinIO storage without having to build their own storage systems from scratch.
MinIO is an AGPL-licensed open source software-defined object storage project that was cofounded by the creator of the open source Gluster file system. Its system is hardware-agnostic, and can run on any commodity infrastructure. But it does provide reference architectures for popular system providers such as Cisco, Dell EMC, Hewlett Packard Enterprise (HPE), and Lenovo. It runs in containers and, as noted below, until now has API-driven with S3 compatibility.
It’s part of a wave of third-party object storage providers that offer their own take on Amazon S3-compatible cloud object storage, providing the building block for replicating the cloud environment without the cloud vendor lock-in. MinIO differentiates its offering through performance levels that it claims are faster than cloud providers. Founded back in 2014, the company claims an installed base running more than 7.7 million instances in AWS, Azure, and GCP, which it states is more than all private cloud object storage instances combined.
Until recently, MinIO was not for the faint of heart – or at least, IT organizations without skills for coding against APIs to build their own private cloud environments or storage infrastructures from scratch. It supported deployment on K8s clusters, but you had to build it yourself with Helm charts, and to monitor it, you had to write code to APIs. The slew of announcements today allows organizations to avoid all that. The K8s implementation is now prepackaged into an operator – this doesn’t mean that the organization won’t need K8s knowledge, but implementation will not be as messy.
Furthermore, there will be a visual control pane built on Grafana, so the organization does not have to mess with Grafana or any other open source visualization to build its own pane of glass. Sand if you want to check the health of your MinIO deployment, against, you won’t have to mess with APIs, as there is a new SUBNET Health service that provides a full view of system diagnostics.
Growing interest in hybrid cloud, where organizations are seeking to replicate the operational simplicity of the cloud control plane for data or workloads that cannot readily run in a public cloud has brought a wave of providers delivering different pieces of software-defined cloud infrastructure. They range from hyperconverged infrastructure that adds a control layer atop commodity hardware and infrastructure software such as application or database management, to components such as storage. MinIO initially emerged as an open source storage project that could deliver the durability and linear scale of cloud object storage, and do so with higher performance than what object storage systems are known for.
Like all the other cloud object storage platforms, MinIO uses APIs to decouple storage state from the application—that is, the application does not have to manage the storage subsystem. That simplifies the application and makes it easier to scale storage – a key reason why modern data warehousing architectures are increasingly separating compute from storage.
Like all the other cloud object storage platforms, MinIO uses APIs to decouple storage state from the application—that is, the application does not have to manage the storage subsystem. That simplifies the application and makes it easier to scale storage – a key reason why modern data warehousing architectures are increasingly separating compute from storage.
K8s and containerization are key to MinIO’s differentiation. While traditional API approaches were sufficient to abstract and scale storage independently of the application, containers make storage management more flexible as they can manage state, while K8s orchestration makes it portable across different hybrid and public cloud platforms. And all the benefits of K8s container orchestration – expanding storage capacity, distributing data and so on – are performed independent of the application or underlying storage media. MinIO surveyed its client base, and not surprisingly, found nearly two thirds of workloads were containerized, and of them, over 40% were orchestrated with K8s.
Much of the impetus for this week’s accessibility announcements was spurred on by MinIO’s partnership with VMware Tanzu announced last fall. The need for the K8s operator became urgent because expecting a customer to build out their own multitenant environment was just too much of an ask – the operator now automates this. While the VMware relationship with MinIO is not exclusive, teaming with VMware required MinIO to make the onramps easier.
